Cristalização da sacarose : estudos experimentais, modelagem matemática e influência de impurezas
Crestani, Carlos Eduardo
MetadataMostrar registro completo
Sugarcane is one of the main Brazilian crops. Studies encompassing unit operations of its industrial processes are old and broad in literature. However, the industrial process itself has changed very slowly over the years, especially concerning sugar production. In this sense, sucrose crystallization studies with industrial application purpose have also an open space to develop themselves. This thesis developed a study of sucrose crystallization encompassing the monitoring of solid and liquid phases; the solubility of sucrose from pure and impure aqueous solutions; the influence of the presence of impurities in critical supersaturation of these solutions and, finally; these results were applied in an experimental study of sucrose cooling crystallization. Inline monitoring of the solid phase was performed by converting the Chord Length Distribution (CLD) which is measured by the laser reflection instrument into Particle Size Distribution (PSD) using Artificial Neural Networks, whereas CLD does not have physical meaning of crystal size and it is of complex interpretation. The correlation coefficient between the calculated and experimental PSD is 0.998. Monitoring of the liquid phase was carried out by refractometry, to which a mathematical model was developed for calculating solution concentration from temperature, refraction index and purity of sugar in solution data (mean deviation between experimental and calculated data is 0.34%). Experimental data of sucrose solubility in pure and impure solutions were obtained and these data were used for evaluation of mathematical methods of solid-liquid equilibrium calculation from literature as well as for the development of two new suggestions for mathematical modeling of equilibrium sucrose solutions containing impurities. The first of these proposals was a modification in Nývlt Equation and resulted in an average deviation between experimental and calculated data of 2.54%. The second proposal aimed the adjustment of interaction parameters of an UNIQUAC based activity coefficient based model, with the insertion of a new fictitious “substance” to represent interaction of impurities in solution and resulted in a mean deviation between experimental and calculated data of 6.25%. Finally, these mathematical models were applied to an experimental study of sucrose cooling seeded crystallization in a pilot scale reactor. Results of these crystallization experiments showed high rates of crystal agglomeration and also some difficulties related to measurement of the laser reflection sensor. In all crystals produced, 69% was agglomerates (practically all crystals higher than 1000 m and 80% of crystals with size between 115 and 256 m). These agglomerates were composed by smaller crystals with mean size of 41.21m, which is the sucrose crystal most susceptible to agglomerate.